Conventionally, a technique is known which compares a reference face image with a plurality of face images and causes information corresponding to a face image similar to the reference face image to be presented. In this technique, there are various methods of comparison between the reference face image and a plurality of face images.
For example, Patent Literature 1 discloses a technique which extracts parts of a face such as eyes and mouth from a face image, extracts a feature amount of the face based on an extraction result, obtains a correlation value between face images based on the extracted feature amounts, and outputs a face image with a maximum correlation value. The technique disclosed in Patent Literature 1 basically uses a whole face as an object to be compared.
On the other hand, Patent Literature 2 discloses a technique which collates a face image part file corresponding to a part specified by a user in a face with face image part files stored in a database, selects a plurality of face image part files in descending order of similarity, and causes a whole face image corresponding to face image part data to be displayed as a search result. The technique disclosed in Patent Literature 2 basically uses only a part of elements included in a face as an object to be compared.